in a given fiscal quarter. Contracts are sold with fluctuating prices from $ 0 to $ 1. They settle at $ 1 per share if a company beats estimates and $ 0 per share if it misses. Thus, the prevailing market price serves as a direct, real-money probability estimate of an earnings upside.
For their research, Rabetti, Shao and Zhang looked at 469 contract settlements covering 383 U. S. companies over six months from September 2025 through February 2026. For each observation, the researchers had the Polymarket closing probability on the day before the earnings announcement, and compared that with the analyst consensus from the Institutional Brokers’ Estimate System and then the actual earnings outcome. The authors also matched the data to information from the Center for Research in Security Prices, which offers historical stock market data, as well as matching it to daily stock return data from Yahoo Finance and to Form 4 insider filings with the Securities and Exchange Commission.
This last data point— insider filings— was a critical placebo test, as we’ ll discuss later.
The researchers asked three questions:
• Do prediction market probabilities outperform analyst consensus in forecasting earnings beats, even after controlling for the analysts’ access to information, weighting for those companies whose lack of information make them more difficult to cover( and for which prediction markets might thus have an upper hand)?
• Is the prediction market’ s informational advantage based on real economic content?
• Why does the prediction market outperform? Which specific distortions in analyst forecasts drive the gap?
Dramatically Outperforming Analyst Consensus
The study found that prediction markets saw a lower gap between the probability of market prices and the actual outcome. When it came to accurately predicting a simple“ and / or” question— whether a company either did hit or did miss its earnings estimates— the prediction markets correctly called the events 78.5 % of the time, while the analyst consensus called it only 43.7 % of the time. That’ s a 35-point gap.
Investment analysts, the study found, were structurally, systematically predicting the wrong outcome for companies: The actual rate at which company earnings beat estimates was 75.9 %, even though the analysts consensus implied that the probability would be only 43.9 %. In other words, if you simply flipped an analyst forecast— betting on a company beating an earnings estimate every time analysts predicted a miss— you would have been right more often than the analysts were. That’ s how badly miscalibrated the consensus was.
In“ horse-race” regressions that pitted prediction market probabilities against analyst consensus directly, the probability market still came out ahead and showed more ability to foresee final results. The authors said that this was true even when the characteristics of certain companies were taken into account( such as their size, analyst coverage and trading volume)— specifically ruling out the possibility that prediction markets were merely winning because they covered easier-to-forecast companies.
Real Economic Content
The study also found that when a prediction market guessed that there would be an earnings surprise, it later anticipated an abnormal realized return in the company’ s stock price.“ Larger [ prediction-market ]- implied surprises are associated with larger announcement returns, suggesting that [ prediction market ] prices capture the magnitude of market reactions.”
The advantage of the prediction market was also larger for non-GAAP reporting firms than for GAAP firms, which is consistent with the idea that these platforms are most valuable where the reporting discretion is greatest and analysts’ findings can suffer from distortion.
Prediction Market Prices Update Continuously
The researchers found that even before companies made their earnings announcements, a gap formed and grew steadily between the prediction market contract prices for those firms predicted to beat their earnings and those predicted to miss, a diver- gence that continued to grow continuously from seven days before the earnings call to the day before. Critically, late-stage price movements predicted outcomes after earlier price levels had been controlled for.
What drove this late-stage updating? The researchers tested whether it reflected corporate insider leakage— illegal tipping— by examining SEC Form 4 insider trading filings. They found no relationship between insider trading activity and prediction market accuracy. Instead, the late-stage price movements reflected the aggregation of legally permissible but hard-to-observe private signals.
Why Does the Prediction Market Win?
This is where the paper makes its most important contribution. The researchers identified four distinct reasons that prediction markets overcome the failures of analyst consensus. 1. Analyst Bias and Incentive Conflicts The academic literature has long documented the ways analysts systematically bias their forecasts. One reason is that they face incentives to issue optimistic forecasts to preserve their access to companies’ management and to support their brokerage relationships. They face backlash and threats to their jobs for estimates that are too far out of the mainstream. Companies, meanwhile, engage in strategic“ walkdowns,” guiding analyst estimates to beatable levels. This herding behavior makes it unlikely that analysts will deviate too sharply from the norm, and thus their estimates are less dispersed.
The result is a consensus that consistently underestimates the probability that a company will beat its earnings.
The prediction market’ s advantage, meanwhile, grew when a company had an extended streak of beating earnings. In fact, when a company beat earnings over longer periods, the prediction markets’ forecasts improved while the analysts’ accuracy deteriorated over the length of the streak. That suggests there’ s an incentivedriven bias, not a fundamentally easier forecasting environment.
The bias was more prevalent in years where companies were issuing equity, a pe-
SEPTEMBER / OCTOBER 2026 | FINANCIAL ADVISOR MAGAZINE | 43